Support for Drupal 7 is ending on 5 January 2025—it’s time to migrate to Drupal 10! Learn about the many benefits of Drupal 10 and find migration tools in our resource center.
When watchdog() is being called now the text is being expanded, so instead of using
Spam: @message
it is using
Spam:
This is causing issues with the mongodb watchdog because it groups all the watchdog entries together by the raw message which generally means that there is only 1 watchdog entry for Spam, Ham, and Unsure instead now with this change the majority of my main watchdog listing page is taken up with spam messages.
Comment | File | Size | Author |
---|---|---|---|
#18 | watchdog_not_being-1556682-18.patch | 4.37 KB | Primsi |
#18 | watchdog_not_being-1556682-18_TEST_ONLY.patch | 1.43 KB | Primsi |
#16 | mollom-1556682.patch | 2.94 KB | Primsi |
#13 | mollom-1556682.patch | 1.23 KB | Primsi |
#5 | mollom-1556682-2.patch | 1.1 KB | Rok Žlender |
Comments
Comment #1
gordon CreditAttribution: gordon commentedI found that in the merging of the watchdog messages it was preprocessing the messages, removing the arguments and making the messages being posted as a single text string.
Ordinarily this would not be a problem, but using 3rd party watchdog tools like mongodb_watchdog which consolidates the log messages based upon the raw message (i.e. the message with the place holders still in tack "spam: %teaser") which makes it harder to manage the messages.
Also the side effect of this with mongodb because it creates a new collection for every new message text which will have ultimately 1 or 2 messages in it, was requiring that mongodb consumed a hugh amount of space.
I have made a change so that when it merges all the watchdog messages it will now merge both the message and the arguments. With the arguments I have put in some conflict detection and arguments will be renamed if there is going to be a conflict.
Comment #3
sunWe've introduced the new formatted log messages in #1000690: Revamp and reduce log messages (was: Not clear that fail-over is working)
The current log messages are always different, because we are grouping multiple requests to Mollom into a single message, but still want to see the full history. Thus, I'm not sure how this could use any kind of always identical placeholders to make Mongo happy...?
Comment #4
gordon CreditAttribution: gordon commentedActually the patch that I have submitted, it gives the consolidated messages, but retains all the arguments so that mongodb can still consolidate, and I don't get a run away in space.
I need to take a look at the patch and see why it failed.
Comment #5
Rok Žlender CreditAttribution: Rok Žlender commentedWe got bitten by this as well mollom create about 21k collections since we rolled new version and are in process of applying this patch to production. There is a bug in the patch $new_arg is not defined I think you meant to use $new_key there attaching fixed up patch.
Comment #6
Rok Žlender CreditAttribution: Rok Žlender commentedComment #8
Rok Žlender CreditAttribution: Rok Žlender commentedComment #9
Rok Žlender CreditAttribution: Rok Žlender commented#5: mollom-1556682-2.patch queued for re-testing.
Comment #10
Rok Žlender CreditAttribution: Rok Žlender commentedBumping this to critical no way to say how many sites out there use MongoDB and Mollom but for those they do this bug can crash their servers.
Comment #11
eshta CreditAttribution: eshta commentedI'm re-evaluating this to see if it is a release candidate (granted some time after initial post).... I'm not understanding the point of removing the call to _mollom_format_string because it is only sanitizing the data.
Comment #12
eshta CreditAttribution: eshta commentedComment #13
Primsi CreditAttribution: Primsi commentedRerolling #5.
Comment #14
eshta CreditAttribution: eshta commented-- changing status so that the patch gets run against testbots.
A small review:
This patch removes the only reference to _mollom_format_string(). The comments should be updated (they refer to it) and the _mollom_format_string() function should be removed.
---------
Also this is lacking any sort of test of the new conflicting arguments functionality.
Comment #16
Primsi CreditAttribution: Primsi commentedRerolled the patch on dev branch and addressed the review issues.
Comment #17
eshta CreditAttribution: eshta commentedThis is great work :-)
Would love to see some more confirmation from folks running MongoDB (will take some extra time for me to get this one up and running to test locally).
Comment #18
Primsi CreditAttribution: Primsi commentedI added a test to the patch. Hope that covers it enough.